ABSTRACT
-
Background/Aims
Hepatocellular carcinoma (HCC) exhibits substantial morphological and biological heterogeneity. Clinical and molecular relevance of the infiltrative subtype remains poorly defined in the context of cancer immunotherapy. We aimed to evaluate the prognostic impact and molecular features of infiltrative HCC in patients treated with first-line atezolizumab plus bevacizumab (Ate/Bev).
-
Methods
We included 307 patients with advanced HCC treated with Ate/Bev and classified them into four gross morphological types based on imaging. Multi-omics profiling was conducted on tumor samples. Type IV infiltrative signature was derived and externally validated using five independent HCC cohorts, including IMbrave150.
-
Results
Infiltrative morphology, encompassing pure and mixed forms, was present in 42.7% of advanced HCC and associated with advanced disease features and compromised liver function. Patients with type IV infiltrative HCC showed lowest objective response rate (14.6%) and worst progression-free (median, 2.8 months) and overall survival (median, 7.1 months). Infiltrative morphology remained an independent predictor of poor outcomes after multivariable adjustment for confounders, including intrahepatic tumor extent. Genomic profiling revealed enriched TP53 and ATM loss-of-function mutations in type IV infiltrative HCC. Transcriptomic and proteomic analyses identified consistent activation of tumor proliferation, epithelial-mesenchymal transition, TGF-β signaling, and immunosuppressive pathways in type IV infiltrative HCC. Type IV infiltrative signature was significantly associated with poor survival across external datasets and retained independent prognostic value.
-
Conclusions
Infiltrative HCC is a clinically aggressive and molecularly distinct subtype of advanced HCC. Morphological classification and type IV infiltrative signatures may guide risk stratification and therapeutic decision-making in advanced HCC treated with immunotherapy.
-
Keywords: Advanced hepatocellular carcinoma; Immunotherapy; Gross classification; Infiltrative morphology; Multiomics analysis
Study Highlights
• Among patients with advanced HCC treated with first-line Ate/Bev, type IV infiltrative HCC had the lowest response rate and worst survival outcomes compared with other subtypes. Infiltrative morphology remained an independent predictor of poor outcome even after adjustment for confounding factors.
• Integrative multi-omics analysis revealed that type IV infiltrative HCC is a molecularly distinct subtype, enriched for TP53 and ATM loss-of-function mutations and characterized by upregulation of cell proliferation, epithelial-mesenchymal transition, TGF-β signaling, and an immunosuppressive tumor microenvironment.
• Type IV infiltrative signature was developed and validated across multiple external cohorts, including IMbrave150.
Graphical Abstract
INTRODUCTION
Hepatocellular carcinoma (HCC) is a biologically and clinically heterogeneous malignancy in which both tumor burden and underlying liver function critically affect prognosis and therapeutic decisions [
1-
5]. HCC classification extends beyond tumor size and number, incorporating features of invasiveness, such as vascular invasion and metastatic potential, which are key prognostic factors that affect treatment outcomes [
6-
8].
Morphologically, HCC can be classified into subtypes, including nodular, massive, and diffuse/infiltrative types, based on gross imaging characteristics [
9,
10]. Among these, infiltrative HCC is distinguished by its diffuse, ill-defined margins, lack of encapsulation, and a high frequency of vascular invasion, particularly portal vein tumor thrombosis (PVTT) [
11-
13]. Infiltrative morphology becomes increasingly prevalent in advanced HCC; however, it is relatively uncommon in early-stage disease [
14,
15]. The insidious growth pattern of infiltrative HCC often makes it difficult to detect by conventional imaging, thereby contributing to delayed diagnosis and poor prognosis [
12,
14,
16]. These clinical features suggest that infiltrative HCC is not simply a morphological variant, but rather a biologically distinct subtype with aggressive tumor behavior [
17,
18]. Considering the clinical significance of infiltrative morphology, the 2022 update of the Barcelona Clinic Liver Cancer (BCLC) staging system introduced it as a subclass within intermediate-stage (BCLC B) HCC [
19,
20]. However, despite its acknowledged importance, the prognostic and biological relevance of infiltrative morphology in the context of cancer immunotherapy remains poorly understood. Given the complex tumor- and liver-related factors that affect HCC prognosis, a comprehensive evaluation that adjusts for key prognostic variables is essential to accurately determine the independent impact of infiltrative morphology on clinical outcomes [
21-
25].
In this study, we systematically classified the gross tumor morphology of patients with advanced HCC treated with first-line atezolizumab plus bevacizumab (Ate/Bev). We aimed to define the clinical impact of infiltrative morphology and identify its underlying molecular features through integrated genomic, transcriptomic, and proteomic analyses, with a particular focus on its association with resistance to first-line Ate/Bev immunotherapy. In addition, the prognostic significance of infiltrative transcriptomic signatures was validated across external datasets.
MATERIALS AND METHODS
Study design and patients
This study enrolled patients with advanced HCC treated with first-line Ate/Bev at three tertiary cancer centers in Korea between June 2020 and July 2023. Eligibility criteria were as follows: 1) individuals aged ≥20 years; 2) advanced HCC confirmed by histologic/cytologic diagnosis or clinical and radiologic features according to the American Association for the Study of Liver Diseases criteria; 3) no prior systemic therapy; 4) Child–Pugh class A or B7; and 5) Eastern Cooperative Oncology Group (ECOG) performance status of 0–1. This study was conducted in accordance with the ethical guidelines of the Declaration of Helsinki and approved by the participating hospitals’ Institutional Review Boards (CHA Bundang Medical Center, CHA-2017-11-052, CHA-2017-11-054, CHA-2020-12-030; Ulsan University Hospital, 2020-12-006; Haeundae Paik Hospital, 2020-12-019-001). Written informed consent was obtained from all the patients.
Gross morphologic classification
Patients with advanced HCC were classified into four gross morphological types based on the classifications of Eggel and Liver Cancer Study Group of Japan [
9,
26]. The definition of each type is as follows: type I, a single round or oval nodule with a well-demarcated margin, without interruption or protrusion at the interface between the tumor and surrounding liver parenchyma; type II, a single nodule with extra-nodular growth; type III, a multinodular confluent tumor formed by multiple nodules that are closely packed together; and type IV, a diffuse infiltrative tumor with a poorly demarcated margin, comprising numerous minute tumor nodules throughout the liver parenchyma. Two board-certified radiologists (C.A. and S.J., with 12 and 15 years of experience in abdominal imaging, respectively) independently evaluated the images and performed morphological classification. Inter-observer agreement for morphological classification, assessed using Cohen’s kappa, yielded a value of 0.74, indicating substantial agreement between the two radiologists. Any discrepancies in the classification were resolved by consensus after discussion.
The Student’s t-test or one-way ANOVA was used to compare continuous variables, and the Fisher’s exact test or the chi-squared test was used to compare categorical variables. Progression-free survival (PFS) and overall survival (OS) were analyzed by constructing the Kaplan–Meier curves using R package survival v3.5-7. The log-rank test was used to evaluate the statistical significance of differences in survival outcomes. Univariate and multivariate Cox proportional hazards regression analyses were conducted to identify the clinical factors and gross HCC types associated with PFS and OS. All statistical analyses were performed at a significance level of P-value <0.05. All plots were generated using R v4.1.3 and ggplot2 v3.4.4.
RESULTS
Clinical characteristics of patients with advanced HCC
This study included 307 patients with advanced HCC treated with first-line Ate/Bev. Baseline clinical characteristics of the patients are presented in
Table 1. Median age was 62.0 years (interquartile range 55.0–69.0 years), and 85.0% of the patients were male. Hepatitis B virus (HBV) infection (64.8%) was the most common etiology of HCC. Baseline α-fetoprotein (AFP) levels of ≥400 ng/mL were present in 41.5% of the patients, and PVTT was observed in 38.8%. Most patients had extrahepatic spread (60.6%) and had received at least one prior local treatment (60.6%) for HCC. Most patients had Child–Pugh class A (76.5%) and BCLC stage C (84.7%) disease.
Workflow of the study is shown in
Figure 1. The gross type of advanced HCC was assessed by radiologists based on the tumor morphology in cross-sectional images. Tumors were classified into four types: type I, a single nodule with smooth, well-defined margins; type II, a single nodule with clear margins and protrusions into the adjacent liver tissue; type III, a lobulated mass from fused nodules with partially indistinct margins; and type IV, a diffuse infiltrative pattern with numerous small nodules dispersed throughout the liver. When multiple types were present, the case was classified based on the most advanced type among them. An overview of the detailed classification of gross types is shown in sunburst plot. Of the 307 patients, 131 (42.7%) had infiltrative type IV nodules, while 38 (12.4%), 72 (23.4%), and 66 (21.5%) had types I, II, and III nodules, respectively, which were observed as single or multiple nodules. Targeted cancer panel sequencing, RNA sequencing, and proteomic profiling were performed on available tumor samples. Furthermore, the genomic, transcriptomic, and proteomic profiles were compared according to the gross type. Finally, the infiltrative genes were further validated using external datasets, including the Ate/Bev arm in IMbrave150 (n=261), The Cancer Genome Atlas (TCGA) cohort (n=370), Villa cohort (n=78), Kim cohort (n=188), and Gao cohort (n=159).
Patients with type IV infiltrative HCC exhibited more advanced disease features compared with other types, including a higher prevalence of extensive intrahepatic tumor involvement (≥50% extent in 34.4% of the patients), elevated serum AFP levels (≥400 ng/mL in 51.1%), and a markedly increased frequency of PVTT (70.2%). Additionally, most patients with type IV tumors were classified as BCLC stage C (93.1%). Infiltrative type IV was also associated with compromised liver function, as reflected by a higher proportion of Child–Pugh class B7 patients (35.9%) and poorer performance status, with 83.2% of the patients having an ECOG performance score of 1. The proportion of patients with prior surgery, transarterial chemoembolization, or radiofrequency ablation was lower in the type IV infiltrative HCC than in the other subgroups.
Prognostic impact of gross tumor type in Ate/Bev-treated advanced HCC
Next, we compared objective response rates (ORR) and survival outcomes of Ate/Bev treatment according to the gross type. Type IV infiltrative HCC had the lowest ORR (14.6%) among all types, whereas type I HCC had the highest ORR (60.5%) (
Table 2). Moreover, the patients with type IV HCC had significantly worse PFS and OS than those with types I, II, or III HCC (median PFS, 2.8 vs. 10.4, 16.0, 4.3 months; median OS, 7.1 vs. not reached [NR], NR, 11.1 months) (
Fig. 2A). Furthermore, irrespective of the intrahepatic tumor extent, type IV HCC consistently showed shorter PFS and OS than the other three gross types (types I–III) (
Fig. 2B). When type IV HCC was further stratified into mixed (IVA) and pure (IVB) forms, both subtypes consistently exhibited worst survival outcomes among all gross HCC types (
Supplementary Fig. 1). Baseline characteristics were largely comparable between the mixed (IVA) and pure (IVB) infiltrative types in most clinical variables (
Supplementary Table 1).
Because various factors could act as confounders of survival outcomes, we further validated the clinical impact of gross HCC type using the multivariable Cox proportional hazards model. After adjusting for ECOG PS, Child–Pugh class, BCLC stage, AFP levels, PVTT, and intrahepatic tumor extent, type IV HCC remained significant predictors of poor PFS and OS (
Fig. 2C and
Supplementary Table 2). Moreover, the unfavorable prognosis of type IV infiltrative HCC remained significant in patients with BCLC stage C (
Supplementary Fig. 2). These findings suggest that type IV infiltrative HCC is an independent predictive indicator of worse treatment response and survival outcomes in patients with advanced HCC treated with Ate/Bev.
To explore the genetic basis of gross morphological types in advanced HCC, we performed targeted cancer panel sequencing of baseline tumor tissues. The most frequent mutations across the entire cohort were in
TP53 (49.1%), followed by
CTNNB1 (21.8%),
ARID2 (12.7%),
ATM (12.7%), and
TERT (12.7%) (
Fig. 3A). Pathway-level analysis revealed substantial variation in the distribution of oncogenic alterations across the gross subtypes. Mutations were clustered in key signaling networks, including p53-RB (56%), Wnt (44%), RTK–RAS–PI3K (25%), chromatin remodeling (22%), and telomerase regulation (13%) (
Fig. 3B). Notably, type IV infiltrative HCC harbored a higher frequency of loss-of-function mutations in
TP53 and
ATM than other gross types of HCC, indicating defective apoptotic control and cell cycle regulation. These findings were consistent with the significant enrichment of p53-RB pathway alterations in type IV HCC (
P=0.04). There were no differences in genomic alterations between the mixed (IVA) and pure (IVB) infiltrative types (
Supplementary Table 3).
We performed RNA sequencing across gross tumor types to delineate the transcriptional basis of the gross morphological subtypes. Type IV infiltrative HCC exhibited extensive differential gene expression compared with noninfiltrative types I–III (
Fig. 4A). Pathway enrichment analysis revealed marked upregulation of gene sets related to cell cycle progression (mitotic spindle assembly, G2/M checkpoint control, and telomere extension) and epithelialmesenchymal transition (EMT), whereas metabolic programs, such as oxidative phosphorylation and adipogenesis, were significantly downregulated in type IV infiltrative HCC (
Fig. 4B and
Supplementary Table 4). Interestingly, the type IV infiltrative HCC tumors were enriched for the Hoshida subclass S1 and macrotrabecular-massive subtype, which is characterized by EMT and transforming growth factor beta (TGF-β) pathway activation. However, the proportion of poorly differentiated tumors did not differ between the infiltrative and non-infiltrative HCC (
Supplementary Fig. 3). Furthermore, immune deconvolution revealed increased infiltration of both effector T cells (Teffs) and regulatory T cells (Tregs) into type IV infiltrative HCC (
Fig. 4C). Consistent with these findings, we further examined key genes associated with the type IV infiltrative type, and type IV infiltrative HCC revealed elevated expression of genes associated with tumor proliferation, pro-tumor cytokines, Tregs, M2-like macrophages, and immune checkpoints compared to types I–III (
Fig. 4D). There were no significant differences in stemness markers such as
CK19,
EPCAM, and
SALL4 expression across gross types. Driver gene expression did not differ consistently between mixed (IVA) and pure (IVB) infiltrative types at the transcriptomic level (
Supplementary Fig. 4).
We performed immunohistochemical (IHC) analysis on available tumor tissues to validate our transcriptome-based findings in infiltrative HCC (
Fig. 4E and
Supplementary Fig. 5). Consistent with multi-omics results, the IHC analysis showed that CD8
+ Teffs and FOXP3
+ Tregs densities were substantially higher in type IV infiltrative HCC than in the other morphological subtypes. However, PD-L1 expression and the number of CD68
+ macrophages did not differ significantly among subgroups. The IHC analysis also demonstrated that Teffs and Tregs were enriched within tumor regions and predominantly co-localized in clustered formations rather than being spatially separated, suggesting potential direct Teff–Treg interactions. In contrast, these cell populations were rarely observed in the periphery, indicating that Teffs had successfully infiltrated the tumor parenchyma but were functionally constrained by the concurrent presence of Tregs. Taken together, these findings provide protein-level validation of our computational deconvolution analysis and add spatial context by demonstrating intratumoral clustering of Teffs and Tregs, thereby reinforcing our interpretation that type IV infiltrative HCC harbors a uniquely immunosuppressive microenvironment.
Because the type IV infiltrative signature was associated with poor clinical outcomes in our Ate/Bev–treated HCC cohort, we further validated its prognostic relevance across multiple external datasets. First, we consistently identified type IV infiltrative signatures in the IMbrave150 study, a global, multicenter, and phase 3 randomized trial evaluating Ate/Bev in patients with advanced HCC. Patients with the type IV infiltrative signature had significantly worse PFS (hazard ratio [HR] 1.76; 95% confidence interval [CI] 1.03–3.02;
P=7.9×10
-3) and OS (HR 1.70; 95% CI 1.00–2.91;
P=0.048) than those with the non-infiltrative signature (
Fig. 5A). These patients also exhibited elevated immune scores for both Teffs and Tregs (
Fig. 5B). Furthermore, patients with the type IV infiltrative signature showed significantly higher Tumor Immune Dysfunction and Exclusion scores, including IFNG, PD-L1, CD8, MDSC, and TAM M2 (
Supplementary Fig. 6).
We further validated the type IV infiltrative signatures across four independent HCC datasets. The type IV infiltrative signature remained significantly associated with poor OS in the TCGA (
P=6.9×10
-3), Villa (
P=8.1×10
-6), Kim (
P=2.7×10
-3), and Gao (
P=0.044) cohorts. Furthermore, the Cox regression analysis revealed that the type IV infiltrative signature was an independent prognostic marker of OS across the four cohorts (
Fig. 5C).
To complement the transcriptomic insights, we conducted a proteomic analysis to characterize protein-level alterations across the gross subtypes. As shown in
Figure 6A, proteomic profiling revealed distinct alterations in differentially expressed proteins in type IV infiltrative HCC compared with non-infiltrative HCC. Gene Ontology enrichment analysis highlighted the upregulation of biological processes linked to DNA replication, homologous recombination, and double-strand break repair, along with the suppression of lipid homeostasis and mitochondrial gene expression (
Fig. 6B). Gene set variation analysis showed significant enrichment of the cell cycle, EMT, TGF-β, and immune-response pathways in type IV infiltrative HCC, consistent with findings from the transcriptomic analysis (
Fig. 6C and
Supplementary Table 5). Notably, post-transcriptional modulation, such as SUMOylation, is more strongly promoted in type IV infiltrative HCC than in other gross types. Collectively, proteomic profiling of type IV infiltrative HCC revealed proliferative and EMT-associated features, which closely mirrored the patterns observed in the transcriptomic analyses.
DISCUSSION
In this study, we provide a comprehensive, multidimensional characterization of the gross morphological subtypes of advanced HCC, with a particular focus on infiltrative HCC. By integrating radiological classification, clinical outcomes, genomic alterations, transcriptomic signatures, and proteomic profiles in a large real-world cohort of patients treated with first-line Ate/Bev, we demonstrated that type IV infiltrative HCC is a biologically distinct and clinically aggressive subtype of advanced HCC. Among all gross types, type IV infiltrative tumors were associated with the lowest ORR, PFS, and OS, and features of advanced tumor burden and compromised liver function. Importantly, type IV infiltrative HCC retained its independent prognostic significance, even after adjusting for established clinical confounders, highlighting the intrinsic aggressiveness of this morphological subtype.
Beyond clinical outcomes, our multi-omics analysis revealed a convergent molecular program in type IV infiltrative tumors, characterized by frequent
TP53 and
ATM mutations and upregulation of the cell cycle and TGF-β signaling, EMT, and immunosuppressive microenvironment signatures (
Fig. 6D). The type IV infiltrative signature was highly concordant and consistently observed across independent external datasets. Notably, the type IV infiltrative signature identified in our cohort was validated as an independent prognostic marker across the IMbrave150 trial and public HCC datasets, thereby supporting its generalizability. Taken together, this study not only establishes the clinical relevance of gross morphology in advanced HCC but also uncovers the molecular basis underlying the aggressiveness and immunotherapy resistance of type IV infiltrative HCC.
In the present study, type IV infiltrative HCC demonstrated consistent upregulation of the TGF-β signaling pathway across transcriptomic and proteomic analyses, suggesting a potential role in driving tumor aggressiveness in advanced HCC. Previous studies have reported that TGF-β plays a dual role in HCC progression, acting as a tumor suppressor during early carcinogenesis, but switching to a tumor promoter at advanced stage [
27]. In the early stage of carcinogenesis, TGF-β is involved in the maintenance of tissue homeostasis by inhibiting tumor growth and protumor inflammation while inducing cell apoptosis [
27,
28]. In contrast, at an advanced stage, TGF-β can acquire the ability to promote cancer cell proliferation and survival [
29,
30]. Moreover, TGF-β facilitates EMT, thereby enhancing tumor invasion and metastasis [
31-
33]. Furthermore, TGF-β has been shown to elicit immunosuppressive phenotypes and recruit Tregs to the tumor microenvironment [
34-
37]. Notably, these tumor-promoting features of TGF-β closely align with the molecular characteristics of type IV infiltrative HCC observed in our study. Therefore, further investigation into the functional role of TGF-β signaling in type IV infiltrative HCC may uncover novel therapeutic vulnerabilities and inform rational combination strategies targeting this aggressive HCC subtype.
The limited efficacy of Ate/Bev in type IV infiltrative HCC likely reflects the convergence of multiple immune evasion mechanisms rather than a single driver pathway. First, in addition to TGF-β signaling, these tumors display upregulation of multiple immune checkpoint molecules, including PD-L1 and CTLA-4, suggesting broad immunoregulatory activation. Second, the persistence of Tregs and immunosuppressive myeloid subsets despite VEGF inhibition indicates that bevacizumab-mediated vascular modulation is insufficient to dismantle the suppressive niche. Third, stromal remodeling, EMT-associated changes, and myeloid reprogramming further reinforce an immune-excluded microenvironment. Together, these multifactorial mechanisms may explain why dual VEGF/PD-L1 blockade provides only partial benefit, underscoring the need for additional strategies to overcome the complex immune resistance of type IV infiltrative HCC.
In the present study, type IV infiltrative HCC was characterized by the enrichment of loss-of-function mutations in
TP53 and
ATM, key tumor suppressors that safeguard genomic integrity. These proteins normally activate DNA repair pathways in response to damage and induce apoptosis when the damage is irreparable, ensuring the elimination of affected cells [
38]. However, mutations in
TP53 and
ATM compromise tumor-suppressive functions, resulting in G1/S checkpoint dysfunction, impaired apoptosis, and heightened genomic instability [
39,
40]. Consequently, DNA-damaged cells develop resistance to apoptosis, which, in turn, fosters the accumulation of additional mutations and drives tumor progression. Collectively, these findings suggest that type IV infiltrative HCC is characterized by profound genomic instability driven by
TP53 and
ATM loss-of-function mutations, a molecular hallmark which may underlie its aggressive clinical behavior and resistance to therapy.
This study has several limitations. Tissue specimens for the molecular analyses were only available for a subset of patients, raising the possibility of a selection bias. Because the multi-omics analysis was performed using a single tumor sample per patient, it may not fully capture the molecular diversity of the entire tumor. HCC is biologically heterogeneous, particularly in large and infiltrative types, which may harbor distinct histological and molecular subregions [
41]. Thus, profiling based on a single specimen could underrepresent the true biological complexity of these tumors. Although the consistent patterns observed across our cohort suggest that key subtype-specific features were captured, future studies using multi-region or spatially resolved profiling, together with circulating biomarkers, will be essential to more comprehensively delineate the heterogeneity of infiltrative HCC. Although the type IV infiltrative signature was externally validated across multiple datasets, further validation of the gross morphological classification, particularly of infiltrative type IV, is warranted in independent external cohorts to confirm its reproducibility and clinical applicability. The classification of gross tumor morphology was based on radiologic interpretation, which may be subject to inter-observer variability, highlighting the need for standardized imaging criteria.
Our study was conducted in an East Asian cohort, where HBV is the predominant etiology. This may limit extrapolation to Western populations, where hepatitis C virus (HCV) and nonalcoholic fatty liver disease (NAFLD) are more common. Because etiology influences both genomic features and the tumor-immune microenvironment, the molecular correlates and prognostic implications of the infiltrative phenotype may differ across populations. For example, HBV-driven HCC is often associated with viral integration and TP53-related genomic instability with chronic antigen-driven T-cell exhaustion, whereas HCV/NAFLD-related HCC more commonly involves Wnt/β-catenin activation and metabolic reprogramming. These etiology-specific programs may differentially shape EMT/TGF-β signaling, immunosuppressive niches, and DNA-damage responses in infiltrative tumors, thereby affecting outcomes and treatment responsiveness. Accordingly, our findings should be interpreted in this context and validated in non-HBV-dominant cohorts.
In recurrent tumors following locoregional therapy, alterations in the tumor microenvironment, such as hypoxia, may contribute to subsequent morphological changes. Therefore, compared with patients without prior locoregional therapies, those who have received such treatments may display distinct infiltrative patterns and transcriptomic/proteomic signatures, which could complicate gross type classification.
Although we identified associations between infiltrative HCC and upregulation of EMT/TGF-β pathways, these findings alone do not establish causality. Functional studies using patient-derived organoid systems or HCC cell lines from infiltrative and nodular tumors are warranted to determine whether activation of these pathways mechanistically drives type IV infiltrative HCC and immunotherapy resistance.
In conclusion, our study demonstrates that gross tumor morphology, particularly type IV infiltrative HCC, has significant prognostic value in advanced HCC treated with Ate/Bev. Through integrative clinical, genomic, transcriptomic, and proteomic analyses, we revealed that type IV infiltrative HCC represents a biologically aggressive and molecularly distinct subtype characterized by proliferative signaling, immune evasion, and a poor immunotherapy response. These findings suggest that infiltrative morphology may aid in clinical decision-making and risk stratification in immunotherapy-treated advanced HCC. Future studies are needed to elucidate the mechanisms of immune resistance and explore novel combination strategies to improve outcomes in this aggressive subtype.
FOOTNOTES
-
Authors’ contributions
Conceptualization: WSL, SW, SHL, SH, CK, HJC; Data curation: WSL, SW, SHL, CA, SJ, GK, HK; Formal analysis: WSL, SW, SHL, GHJ, DYL, JP; Funding acquisition: SH, CK, HJC; Investigation: WSL, SW, SHL, GHJ, Ilhwan Kim, HK, BK, JSK, HYL, CA, SJ, GK, HK, Incheon Kang, HY, SJK, DS, DJS, WYK, DYL, JSL, JP, YK; Methodology: WSL, SW, SHL, SH, CK, HJC; Resources: SH, CK, HJC; Writing – original draft: WSL, SW, SHL, SH, CK, HJC; Writing – review & editing: SH, CK, HJC. All authors approved the final version of the manuscript.
-
Acknowledgements
This research was funded by National Research Foundation of Korea (NRF) grants supported by the Korean government (MSIT) (grant numbers: NRF-2023R1A2C2004339 and RS-2023-NR076871). This research was supported by a grant of the Korea Health Technology R&D project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: RS-2024-00438843). This research was supported by Basic Science Research Program through NRF funded by the Ministry of Education (grant number: RS-2019-NR040073). This research was supported by a research grant from BORYUNG Co.
-
Data availability statement
The type IV infiltrative signature gene list has been provided in Supplementary Table 6. The Bayesian compound covariate predictor model code and processed expression matrices will be made available upon reasonable request to the corresponding author.
-
Conflicts of Interest
Hong Jae Chon holds consulting or advisory roles with Eisai, Roche, Bayer, ONO, MSD, BMS, Celgene, Sanofi, Servier, AstraZeneca, SillaJen, Menarini, and GreenCross Cell, and has received research grants from Roche, Dong-A ST, and Boryung Pharmaceuticals. Chan Kim is involved in consulting or advisory capacities with Roche, ONO, MSD, BMS, Oncocross, and Virocure, and has received research funding from Boryung Pharmaceuticals, Oncocross, SillaJen, and Virocure. The remaining authors declare no potential conflicts of interest
SUPPLEMENTARY MATERIAL
Supplementary material is available at Clinical and Molecular Hepatology website (
http://www.e-cmh.org).
Supplementary Figure 1.
Survival analysis of gross type with progression-free survival (PFS) and overall survival (OS) in patients with advanced hepatocellular carcinoma (HCC) treated with atezolizumab and bevacizumab combination therapy. Kaplan–Meier curves comparing 307 patients with the gross type of HCC based on tumor extent: <25%, 25–50%, 50–75%, and ≥75%. Type IVA refers to HCC in which the gross pattern of type IV is mixed with components of types I, II, or III, whereas type IVB refers to lesions that have a pure gross morphology of type IV.
Supplementary Figure 2.
Prognostic impact of gross type in patients with BCLC stage C. (A) Kaplan–Meier curve of PFS and OS among patients with BCLC stage C (n=260), stratified by gross type (I–IV). (B) Kaplan–Meier curves for PFS and OS according to gross type within subgroups defined by intrahepatic tumor extent (<25%, 25–50%, 50–75%, and ≥75%). Log-rank P-values are shown in each panel.
Supplementary Figure 3.
Hoshida subclass and histological subtype distribution by gross type and infiltrative status. (A) Stacked bar plots showing the proportions of Hoshida subclass S1–S3 and not significant (NS) calls across gross types I–IV. (B) Distribution in non-infiltrative (types I–III) and infiltrative (type IV) tumors. (C) Distribution of morphological subtype and histological differentiation in HCC. Numbers inside the bars indicate number (%). Subclasses were assigned using the nearest template prediction on a log2 (CPM+1) matrix, z-score scaled per gene (n=61). Samples assigned ‘not significant’ denote those without a significant call (P≥0.05). The P-values were calculated using the Cochran-Armitage trend test for S1 versus all other subclasses. MTM, macrotrabecular-massive.
Supplementary Figure 4.
Molecular characteristics of mixed (IVA) and pure (IVB) infiltrative HCC. Shown are box plots of (A) immune scores of effector T cells (Teffs) and regulatory T cells (Tregs), (B) GSVA scores for representative hallmark pathways, and (C) normalized expression of selected key genes across infiltrative subtypes. Statistical comparisons were performed using Student’s t-test for panels A and C, and limma for panel B. GSVA, gene-set variation analysis; ns, not significant.
Supplementary Figure 5.
Immunohistochemical analysis of tumor tissues across gross types. Representative images and comparisons of (A) PD-L1 and (B) CD68 within tumor tissues. P-values were calculated using the Student’s t-test in infiltrative (type IV) vs. non-infiltrative (types I–III) types. Scale bars=60 μm.
Supplementary Figure 6.
Comparison of TIDE scores according to gross types. Prediction of T cell dysfunction and exclusion scores in IMbrave150 using immunosuppressive cell signatures.
Supplementary Table 1.
Baseline characteristics of patients with advanced hepatocellular carcinoma, categorized by mixed (IVA) and pure (IVB) infiltrative types
Supplementary Table 2.
The results of univariable and multivariable Cox regression of the gross type and clinical factors in patients with advanced hepatocellular carcinom
Figure 1.Workflow of the study with gross type in advanced HCC. Representative radiologic imaging and corresponding schematic diagrams showing the distinct gross types observed in patients with advanced HCC. The red arrow shows portal vein thrombosis. Sunburst plot illustrating the classification of gross types of HCC. The outer ring represents the major gross types, while the inner ring shows more detailed types, including both pure and mixed forms. Some icons were created using BioRender.com. Ate/Bev, atezolizumab plus bevacizumab; DEG, differentially expressed gene; DEP, differentially expressed protein; GO, Gene Ontology; HCC, hepatocellular carcinoma; TCGA, The Cancer Genome Atlas; VAF, variant allele frequency.
Figure 2.Survival analysis of gross type with PFS and OS in patients with advanced HCC treated with atezolizumab and bevacizumab combination therapy. (A, B) Kaplan–Meier curves comparing 307 patients with the gross type of HCC based on tumor extent: <25%, 25–50%, 50–75%, and ≥75%. (C) Results of the multivariable Cox regression analysis with PFS and OS of the gross type and clinical factors in 306 patients with advanced HCC. A patient without α-fetoprotein was excluded from the Cox regression analysis. Clinical factors with a P-value <0.05 in the univariable analysis were included. AFP, alpha-fetoprotein; BCLC, Barcelona Clinic Liver Cancer; ECOG PS, Eastern Cooperative Oncology Group performance status; OS, overall survival; PFS, progression-free survival; PVTT, portal vein tumor thrombosis.
Figure 3.Genomic landscape according to gross types. (A) Frequency of genomic alterations of 20 key genes in patients with advanced HCC, including non-synonymous mutations and copy number variations, ranked by their prevalence. (B) Pathway-level representation of genomic alterations across different gross types. The diagram illustrates the proportion of samples harboring alterations in key oncogenic pathways: p53–RB, RTK–RAS–PI3K, telomerase, WNT, and chromatin modifiers. Each gene is annotated with four percentages corresponding to the frequency of activation (red) or inactivation (blue) across the gross types of advanced HCC. HCC, hepatocellular carcinoma.
Figure 4.Transcriptomic characteristics according to gross types. (A) Volcano plot showing differentially expressed genes between infiltrative (type IV) and non-infiltrative (types I, II, and III) types in patients with advanced HCC. (B) Heatmap illustrating distinct hallmark gene sets and cell cycle-related Reactome pathways identified by gene set variation analysis (GSVA), with colors representing scaled mean enrichment scores for each gross type. (C) Box plots showing immune scores of effector T cells (Teffs) and regulatory T cells (Tregs) across gross types. Statistical significance in panel B was calculated using the Student’s t-test comparing each type to all others, whereas panel C was analyzed using limma; *P<0.05. (D) Box plots comparing representative genes associated with tumor proliferation, pro-tumor cytokines, T cell/Tregs/M2-like, immune checkpoints and metabolism across gross types. (E) Immunohistochemical analysis of tumor tissues across gross types. Representative images and comparisons of CD8 and FOXP3 within tumor tissues. P-values were calculated using the Student’s t-test in infiltrative (type IV) vs. non-infiltrative (types I–III) types; *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. Scale bars=60 μm.
Figure 5.Validation of the type IV infiltrative signature of HCC in public datasets using the Bayesian compound covariate predictor model. (A) Kaplan–Meier curves for PFS and OS comparing patients having HCC with predicted infiltrative versus non-infiltrative types in the IMbrave150 cohort (n=261) treated with Ate/Bev. (B) Box plots showing immune scores of Teffs and Tregs in predicted infiltrative versus non-infiltrative types in the IMbrave150 cohort. (C) Kaplan–Meier curves for OS comparing infiltrative and non-infiltrative subtypes in four public HCC datasets (n=795): the TCGA cohort, Villa et al.,S6 2016, Kim et al.,S7 2014, and Gao et al.,S8 2019. Hazard ratios (HRs) were estimated with univariable Cox proportional-hazards regression, and P-values were obtained from two-sided log-rank tests. Ate/Bev, atezolizumab plus bevacizumab; HCC, hepatocellular carcinoma; OS, overall survival; PFS, progression-free survival; Teffs, effector T cells; Tregs, regulatory T cells; TCGA, The Cancer Genome Atlas.
Figure 6.Proteomic characteristics of gross type with advanced HCC. (A) Volcano plot showing differentially expressed proteins (DEPs) between infiltrative (gross type IV) and non-infiltrative (gross types I, II, and III) types in patients with advanced HCC. (B) Bar chart of Gene Ontology (GO) enrichment for DEPs from (A); bars colored by adjusted P-value. (C) Heatmap of GSVA scores for hallmark, KEGG and Reactome gene sets that differ significantly between infiltrative and non-infiltrative HCC. (D) Graphical abstract of the characteristics according to gross classification. Some icons were created using BioRender.com. P-values in panels A and C were calculated using the limma package with a significance threshold of P<0.05. DEPs, differentially expressed proteins; EMT, epithelial-mesenchymal transition; HCC, hepatocellular carcinoma; GSVA, gene set variation analysis; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; TGF-β, transforming growth factor beta; Tregs, regulatory T cells.
Table 1.Baseline characteristics of patients with advanced hepatocellular carcinoma who received atezolizumab plus bevacizumab, categorized by gross type
Table 1.
|
Characteristics |
Overall (n=307) |
Type I (n=38) |
Type II (n=72) |
Type III (n=66) |
Type IV (n=131) |
P-value |
|
Median age |
62 (55–69) |
62 (55–69) |
63 (59–70) |
65 (55–72) |
60 (51–66) |
0.01 |
|
Sex |
|
|
|
|
|
0.30 |
|
Female |
46 (15.0) |
3 (7.9) |
11 (15.3) |
14 (21.2) |
18 (13.7) |
|
|
Male |
261 (85.0) |
35 (92.1) |
61 (84.7) |
52 (78.8) |
113 (86.3) |
|
|
ECOG PS |
|
|
|
|
|
3.0×10-9
|
|
0 |
99 (32.2) |
24 (63.2) |
36 (50.0) |
17 (25.8) |
22 (16.8) |
|
|
1 |
208 (67.8) |
14 (36.8) |
36 (50.0) |
49 (74.2) |
109 (83.2) |
|
|
Etiology |
|
|
|
|
|
0.07 |
|
Hepatitis B |
199 (64.8) |
22 (57.9) |
44 (61.1) |
36 (54.5) |
97 (74.1) |
|
|
Hepatitis C |
23 (7.5) |
5 (13.2) |
7 (9.7) |
4 (6.1) |
7 (5.3) |
|
|
Alcohol |
46 (15.0) |
7 (18.4) |
9 (12.5) |
12 (18.2) |
18 (13.7) |
|
|
MASLD |
39 (12.7) |
4 (10.5) |
12 (16.7) |
14 (21.2) |
9 (6.9) |
|
|
Child–Pugh |
|
|
|
|
|
1.3×10-6
|
|
A |
235 (76.5) |
37 (97.4) |
66 (91.7) |
48 (72.7) |
84 (64.1) |
|
|
B7 |
72 (23.5) |
1 (2.6) |
6 (8.3) |
18 (27.3) |
47 (35.9) |
|
|
BCLC stage |
|
|
|
|
|
1.8×10-5
|
|
B |
47 (15.3) |
15 (39.5) |
13 (18.1) |
10 (15.2) |
9 (6.9) |
|
|
C |
260 (84.7) |
23 (60.5) |
59 (81.9) |
56 (84.8) |
122 (93.1) |
|
|
PVTT |
119 (38.8) |
2 (5.3) |
8 (11.1) |
17 (25.8) |
92 (70.2) |
<2.2×10-16
|
|
Extrahepatic spread |
|
|
|
|
|
0.28 |
|
Absent |
121 (39.4) |
19 (50.0) |
24 (33.3) |
23 (34.8) |
55 (42.0) |
|
|
Present |
186 (60.6) |
19 (50.0) |
48 (66.7) |
43 (65.2) |
76 (58.0) |
|
|
AFP ≥400 ng/mL |
127 (41.5) |
9 (23.7) |
16 (22.2) |
35 (53.0) |
67 (51.1) |
1.1×10-5
|
|
Intrahepatic tumor extent |
|
|
|
|
|
<2.2×10-16
|
|
<25% |
137 (44.6) |
34 (89.5) |
54 (75.0) |
11 (16.7) |
38 (29.0) |
|
|
25–50% |
95 (31.0) |
4 (10.5) |
11 (15.3) |
32 (48.5) |
48 (36.6) |
|
|
50–75% |
60 (19.5) |
0 (0.0) |
7 (9.7) |
16 (24.2) |
37 (28.3) |
|
|
≥75% |
15 (4.9) |
0 (0.0) |
0 (0.0) |
7 (10.6) |
8 (6.1) |
|
|
Prior local treatment |
|
|
|
|
|
|
|
Surgery |
56 (18.2) |
17 (44.7) |
12 (16.7) |
7 (10.6) |
20 (15.3) |
8.8×10-5
|
|
Radiotherapy |
84 (27.4) |
12 (31.6) |
24 (33.3) |
15 (22.7) |
33 (25.2) |
0.45 |
|
TACE |
154 (50.2) |
31 (81.6) |
48 (66.7) |
28 (42.4) |
47 (35.9) |
1.1×10-7
|
|
RFA |
26 (8.5) |
8 (21.1) |
11 (15.3) |
2 (3.0) |
5 (3.8) |
3.9×10-4
|
Table 2.Best overall response to atezolizumab plus bevacizumab
Table 2.
|
Response |
Overall (n=307) |
Type I (n=38) |
Type II (n=72) |
Type III (n=66) |
Type IV (n=131) |
|
Complete response |
9 (2.9) |
3 (7.9) |
4 (5.6) |
1 (1.5) |
1 (0.8) |
|
Partial response |
87 (28.3) |
20 (52.6) |
33 (45.8) |
17 (25.8) |
17 (13.0) |
|
Stable disease |
131 (42.7) |
14 (36.9) |
28 (38.9) |
28 (42.4) |
61 (46.5) |
|
Progressive disease |
67 (21.8) |
1 (2.6) |
5 (6.9) |
17 (25.8) |
44 (33.6) |
|
Not evaluable |
13 (4.2) |
0 (0.0) |
2 (2.8) |
3 (4.5) |
8 (6.1) |
|
ORR |
96 (32.7) (27.5–38.2) |
23 (60.5) (44.7–74.4) |
37 (52.9) (41.3–64.1) |
18 (28.6) (18.9–40.7) |
18 (14.6) (9.5–21.9) |
|
DCR |
227 (77.2) (72.1–81.6) |
37 (97.4) (86.5–99.5) |
65 (92.9) (84.3–96.9) |
46 (73.0) (61.0–82.4) |
79 (64.2) (55.4–72.1) |
Abbreviations
atezolizumab plus bevacizumab
Barcelona Clinic Liver Cancer
differentially expressed proteins
Eastern Cooperative Oncology Group
epithelial-mesenchymal transition
gene set variation analysis
nonalcoholic fatty liver disease
progression-free survival
portal vein tumor thrombosis
transforming growth factor beta
REFERENCES
- 1. Dong LQ, Peng LH, Ma LJ, Liu DB, Zhang S, Luo SZ, et al. Heterogeneous immunogenomic features and distinct escape mechanisms in multifocal hepatocellular carcinoma. J Hepatol 2020;72:896-908.
- 2. Forner A, Llovet JM, Bruix J. Hepatocellular carcinoma. Lancet 2012;379:1245-1255.
- 3. Cheon J, Kim H, Kim HS, Kim CG, Kim I, Kang B, et al. Atezolizumab plus bevacizumab in patients with child-Pugh B advanced hepatocellular carcinoma. Ther Adv Med Oncol 2023;15:17588359221148541.
- 4. Yang H, Kang B, Ha Y, Lee SH, Kim I, Kim H, et al. High serum IL-6 correlates with reduced clinical benefit of atezolizumab and bevacizumab in unresectable hepatocellular carcinoma. JHEP Rep 2023;5:100672.
- 5. Jung WJ, Jo EJ, Kim YJ, Park M, Kim E, Jung YS, et al. Characteristics of immune checkpoint inhibitor (ICI)-related hepatotoxicity based on the baseline liver function. Cancer Res Treat 2025.
- 6. Cha SW, Sohn JH, Kim SH, Kim YT, Kang SH, Cho MY, et al. Interaction between the tumor microenvironment and resection margin in different gross types of hepatocellular carcinoma. J Gastroenterol Hepatol 2020;35:648-653.
- 7. Dendy MS, Camacho JC, Ludwig JM, Krasinskas AM, Knechtle SJ, Kim HS. Infiltrative hepatocellular carcinoma with portal vein tumor thrombosis treated with a single high-dose Y90 radioembolization and subsequent liver transplantation without a recurrence. Transplant Direct 2017;3:e206.
- 8. Choi SJ, Chung SW, Choi J, Kim KM, Kim HD, Yoo C, et al. Clinical and radiologic predictors of response to atezolizumabbevacizumab in advanced hepatocellular carcinoma. Cancer Res Treat 2024;56:1219-1230.
- 9. Madani SP, Mirza-Aghazadeh-Attari M, Mohseni A, Pawlik T, Kamel IR. Diffuse infiltrative hepatocellular carcinoma: Multimodality imaging manifestations. J Surg Oncol 2023;127:385-393.
- 10. Yopp AC, Mokdad A, Zhu H, Mansour JC, Balch GC, Choti MA, et al. Infiltrative hepatocellular carcinoma: natural history and comparison with multifocal, nodular hepatocellular carcinoma. Ann Surg Oncol 2015;22 Suppl 3:S1075-1082.
- 11. Mehta N, Fidelman N, Sarkar M, Yao FY. Factors associated with outcomes and response to therapy in patients with infiltrative hepatocellular carcinoma. Clin Gastroenterol Hepatol 2013;11:572-578.
- 12. Reynolds AR, Furlan A, Fetzer DT, Sasatomi E, Borhani AA, Heller MT, et al. Infiltrative hepatocellular carcinoma: what radiologists need to know. Radiographics 2015;35:371-386.
- 13. Benvegnù L, Noventa F, Bernardinello E, Pontisso P, Gatta A, Alberti A. Evidence for an association between the aetiology of cirrhosis and pattern of hepatocellular carcinoma development. Gut 2001;48:110-115.
- 14. Kneuertz PJ, Demirjian A, Firoozmand A, Corona-Villalobos C, Bhagat N, Herman J, et al. Diffuse infiltrative hepatocellular carcinoma: assessment of presentation, treatment, and outcomes. Ann Surg Oncol 2012;19:2897-2907.
- 15. Fan Z, Jin M, Zhang L, Wang N, Li M, Wang C, et al. From clinical variables to multiomics analysis: a margin morphology-based gross classification system for hepatocellular carcinoma stratification. Gut 2023;72:2149-2163.
- 16. Bruix J, Sherman M. Management of hepatocellular carcinoma: an update. Hepatology 2011;53:1020-1022.
- 17. Chiang CM, Huang KK, Lee CT, Hong TC, Wu JS, Wu HT, et al. Comparing immunotherapy effectiveness for unresectable hepatocellular carcinoma: infiltrative versus noninfiltrative types in real-world settings. Ther Adv Med Oncol 2025;17:17588359241312141.
- 18. He J, Shi J, Fu X, Mao L, Zhou T, Qiu Y, et al. The clinicopathologic and prognostic significance of gross classification on solitary hepatocellular carcinoma after hepatectomy. Medicine (Baltimore) 2015;94:e1331.
- 19. Kim SH, Kim JH, Kim GH, Kim JH, Ko HK, Chu HH, et al. Intermediate-stage (BCLC stage B) infiltrative hepatocellular carcinoma: safety and efficacy of chemoembolization. Eur Radiol 2023;33:8736-8744.
- 20. Reig M, Forner A, Rimola J, Ferrer-Fàbrega J, Burrel M, Garcia-Criado Á, et al. BCLC strategy for prognosis prediction and treatment recommendation: The 2022 update. J Hepatol 2022;76:681-693.
- 21. Giannini EG, Bucci L, Garuti F, Brunacci M, Lenzi B, Valente M, et al. Patients with advanced hepatocellular carcinoma need a personalized management: A lesson from clinical practice. Hepatology 2018;67:1784-1796.
- 22. Demirjian A, Peng P, Geschwind JF, Cosgrove D, Schutz J, Kamel IR, et al. Infiltrating hepatocellular carcinoma: seeing the tree through the forest. J Gastrointest Surg 2011;15:2089-2097.
- 23. Rhee H, Chung T, Yoo JE, Nahm JH, Woo HY, Choi GH, et al. Gross type of hepatocellular carcinoma reflects the tumor hypoxia, fibrosis, and stemness-related marker expression. Hepatol Int 2020;14:239-248.
- 24. Hui AM, Takayama T, Sano K, Kubota K, Akahane M, Ohtomo K, et al. Predictive value of gross classification of hepatocellular carcinoma on recurrence and survival after hepatectomy. J Hepatol 2000;33:975-979.
- 25. Kim Y, Kim JS, Kang B, Kim I, Kim H, Lee WS, et al. Distinct characteristics and changes in liver function of patients with hepatocellular carcinoma treated with atezolizumab plus bevacizumab for more than 1 year. Cancer Res Treat 2024;56:1231-1239.
- 26. Kudo M, Kawamura Y, Hasegawa K, Tateishi R, Kariyama K, Shiina S, et al. Management of hepatocellular carcinoma in Japan: JSH consensus statements and recommendations 2021 update. Liver Cancer 2021;10:181-223.
- 27. Massagué J, Sheppard D. TGF-β signaling in health and disease. Cell 2023;186:4007-4037.
- 28. Gonzalez-Sanchez E, Vaquero J, Férnandez-Barrena MG, Lasarte JJ, Avila MA, Sarobe P, et al. The TGF-β pathway: a pharmacological target in hepatocellular carcinoma? Cancers (Basel) 2021;13.
- 29. Soukupova J, Malfettone A, Hyroššová P, Hernández-Alvarez MI, Peñuelas-Haro I, Bertran E, et al. Role of the transforming growth factor-β in regulating hepatocellular carcinoma oxidative metabolism. Sci Rep 2017;7:12486.
- 30. Zhang K, Zhang M, Luo Z, Wen Z, Yan X. The dichotomous role of TGF-β in controlling liver cancer cell survival and proliferation. J Genet Genomics 2020;47:497-512.
- 31. Xu J, Lamouille S, Derynck R. TGF-beta-induced epithelial to mesenchymal transition. Cell Res 2009;19:156-172.
- 32. Heldin CH, Vanlandewijck M, Moustakas A. Regulation of EMT by TGFβ in cancer. FEBS Lett 2012;586:1959-1970.
- 33. Hao Y, Baker D, Ten Dijke P. TGF-β-mediated epithelial-mesenchymal transition and cancer metastasis. Int J Mol Sci 2019;20.
- 34. Chen J, Gingold JA, Su X. Immunomodulatory TGF-β signaling in hepatocellular carcinoma. Trends Mol Med 2019;25:1010-1023.
- 35. Batlle E, Massagué J. Transforming growth factor-β signaling in immunity and cancer. Immunity 2019;50:924-940.
- 36. Xin X, Cheng X, Zeng F, Xu Q, Hou L. The role of TGF-β/SMAD signaling in hepatocellular carcinoma: from mechanism to therapy and prognosis. Int J Biol Sci 2024;20:1436-1451.
- 37. Lee WS, Yang H, Chon HJ, Kim C. Combination of anti-angiogenic therapy and immune checkpoint blockade normalizes vascular-immune crosstalk to potentiate cancer immunity. Exp Mol Med 2020;52:1475-1485.
- 38. Karakostis K, Malbert-Colas L, Thermou A, Vojtesek B, Fåhraeus R. The DNA damage sensor ATM kinase interacts with the p53 mRNA and guides the DNA damage response pathway. Mol Cancer 2024;23:21.
- 39. Chen X, Zhang T, Su W, Dou Z, Zhao D, Jin X, et al. Mutant p53 in cancer: from molecular mechanism to therapeutic modulation. Cell Death Dis 2022;13:974.
- 40. Choi M, Kipps T, Kurzrock R. ATM mutations in cancer: therapeutic implications. Mol Cancer Ther 2016;15:1781-1791.
- 41. Zhu S, Hoshida Y. Molecular heterogeneity in hepatocellular carcinoma. Hepat Oncol 2018;5:Hep10.
, Chan Kim1
